Samsung makes 64 Gb NAND flash chip

[jtr1962]

Samsung Electronics revealed that they have made a prototype 64 Gb NAND flash chip using a 30 nm process. Production is expected to start in 2009. The new process promises to improve cost efficiency as well as increase capacity to 8 GB on a single chip. I remember back when 8 GB represented state of the art in a 3.5" mechanical hard disk, and it doesn't seem all that long ago. As far as I know, the largest currently shipping flash NAND chip is 16 Gb (2 GB). Basically, this development looks to increase the capacity of everything flash-based by a factor of four, while potentially reducing the cost. I guess then I'll be able to get a nice 16 GB USB drive for $30 or so sometime in 2009, with sizes up to 64 GB available for those willing to spend more.

On another note, sixteen of these on a double-sided DIMM equals 128 GB of non-volatile storage. Not bad.

 

[udaman]

Samsung Electronics revealed that they have made a prototype 64 Gb NAND flash chip using a 30 nm process. Production is expected to start in 2009. The new process promises to improve cost efficiency as well as increase capacity to 8 GB on a single chip. I remember back when 8 GB represented state of the art in a 3.5" mechanical hard disk, and it doesn't seem all that long ago. As far as I know, the largest currently shipping flash NAND chip is 16 Gb (2 GB). Basically, this development looks to increase the capacity of everything flash-based by a factor of four, while potentially reducing the cost. I guess then I'll be able to get a nice 16 GB USB drive for $30 or so sometime in 2009, with sizes up to 64 GB available for those willing to spend more.

On another note, sixteen of these on a double-sided DIMM equals 128 GB of non-volatile storage. Not bad.

From the PR:

SaDPT represents a pivotal advancement beyond the charge trap flash (CTF) technology that Samsung developed for NAND flash last year when it introduced a new material (silicon nitride) and a new structural configuration. SaDPT resolves a critical bottleneck to forming sub-30nm circuitry by expanding the role that conventional lithography technology plays in the manufacturing process. Both Samsung’s CTF-based NAND flash technology and SaDPT are expected to provide improvements in cost efficiency for next-generation nanometer-scale designs.
Samsung’s SaDPT will employ existing photolithography equipment in 30nm-class production, which is expected to be commercialized beginning in 2009. By utilizing conventional photolithography equipment, Samsung can not only significantly speed up the process but also improve the cost efficiency of its manufacturing operations without additional facility investment. Samsung has applied for 30 patents in connection with its new 64Gb flash device.
Samsung also has developed a 32Gb single level cell (SLC) NAND flash chip based on the same technology applied to its 64Gb chip. Samsung’s continued success in introducing higher density NAND flash will intensify demand for solid state drives in notebooks, and for other NAND-based storage devices in applications such as digital camcorders and enterprise servers.

Well Samsung has propensity (just like Cree ) for hyping up new technology that may or may not be just all that , along with taking much longer than expected to bring it to market. The CTF they annouced last year based on 40nm process, where is it? They were supposed to have up to 128GB CTF 40nm NAND out by the end of this year, yet AFAIK, they have yet to ship a single unit of any size of CTF technology, so I won't hold my breath or count on any 128GB 30nm, even newer technological developments to come online in 2009.

Not really sure by the vague PR, but what exactly is more privotal about SaDPT over CTF, which last year Samsung was saying could scale down to less than 20nm.

MLC are at present slower performing as compared to SLC, so I would be more interested in seeing what develops with the SaDPT SLC 32Gb chips...oh somewhere around 2010 if we are lucky ...right about the time Cree ships in commercially available quantities, not just prototype 128lm/w white LED's, that they just annouced in a September PR.

128GB in 2009? I think 7.2k rpm HD's based on even higher densities than on the current 320GB Samsung 5.4k 2.5in drives now being sold for Alienware/Dell laptops, look more compelling for cost consideration, all except perhaps those who need 15k drive performance and are willing to pay that price premium... kind of like all those people who buy the more expensive Apple computers...and btw see my other thread on that

 

[jtr1962]

Well Samsung has propensity (just like Cree ) for hyping up new technology that may or may not be just all that , along with taking much longer than expected to bring it to market.

Actually, isn't that Lumileds you're thinking of? They're the ones who hyped their K2 for months before it was available. When it finally was, it turned out to be a bust. Cree usually delivers production product within 18 months after it first appears in the lab.

MLC are at present slower performing as compared to SLC, so I would be more interested in seeing what develops with the SaDPT SLC 32Gb chips..

Same here. SLC are also more reliable than MLC. Remember that to get two bits on a cell instead of one you need to be able to store and read four different levels. Four bits requires 16 levels, etc. In other words, linear increases in density require exponential increases in complexity. Not worth it IMO.

 

[udaman]

jtr's link is now registration only, so here's the other one, with more revealing advertising .

http://www.engadget.com/2007/10/23/samsung-reveals-first-30-nm-64gb-nand-chip-and-a-bit-of-skin/


Don't you just love all the vaporwave hype between the two leading contenders?

Now we have Toshiba with "about" 100Gb potential (huh, why not 128Gb?) on an even smaller "potential" (means they think they can do it) 10nm process, lol. Only four generations away from full production, mass produced lol. Let me see here, if it takes Samsung "About" 2 years before then get 50nm larger capacity/faster drives (controllers seem to be what make the most difference, not cell tech) are available at really hight prices, and if it takes how long for Toshiba to put out their 128GB NAND flash tech using 56nm process, extrpolating to 10nm & 4 generations to get a approixmate 5 fold increase in capacity (assuming they can ever actually get such technology into mass production before some other development occurs)...should be here in...oh, say 4 gens x 2yrs lead time = 8yrs, lol.

Toshiba lays groundwork for 100Gb flash

http://www.electronista.com/articles/07/12/12/toshiba.100gb.flash.test/

In recent years, flash memory generations have advanced roughly once per year and doubled capacity in each case.

Sure they do, paper generational launches every year, actually annoucment to real production availability in retail channels = ~2yrs.

If you ask me, 4-8yrs from now is a long, long time; and by then even 1-2GB NAND/SSD capacity will probably seem just as puny as 32 or 64GB does now. However, USB 2.0 or 3.0 distribution of OS and other mega-bloatware on thumbnail drives, seems imminent. Apple is supposedly going to unveil an ultrathin laptop using SSD drive and *no* optical drive...got to be someway to get software in that system other than by wireless means.

 

[ddrueding]

...got to be someway to get software in that system other than by wireless means.

Why?

 

[jtr1962]

If you ask me, 4-8yrs from now is a long, long time; and by then even 1-2GB NAND/SSD capacity will probably seem just as puny as 32 or 64GB does now.

I think you meant 1-2TB. :tounge: A lot of whether or not SSDs take over in the next few years, besides cost per GB, is indeed capacity. I really don't see magnetic drives increasing capacity by leads and bounds. It seemed like we were stuck at 500GB for ages before the bumps to 750GB and finally 1TB. I'd guess if we're lucky in 5 years magnetic disks will be up 3 or 4 TB. Now if SSDs advance 4 generations by then (that's roughly 2 or 16 times increase in capacity), we should be up to around 2TB. No idea what it will cost, but I'd guess with economies of mass production probably a few hundred dollars-about the same as the highest capacity hard disks. SSDs may end up advancing faster or slower. Hard to predict these things. LEDs weren't supposed to reach 100 lm/W in production until 2010 or so but we're there now. Perceived demand for something has a lot to do with how much R&D is thrown at it. Both LEDs and SSDs are hot, so to speak. I'm sure soon Intel might be into making flash chips, seeing that demand is so hot. I think time will prove your or even my more optimistic estimates to be quite pessimistic. All I know is when SSDs get to around $99 for at least 250 GB I'm in, at least for a boot drive. Once they get to a TB or more at that price, I'm not even looking at mechanical disks any more. I have little doubt my next system build in perhaps 3 or 4 years will incorporate an SSD in some way. If I'm lucky they may even become affordable/large enough to put in my current system.

 

[P5-133XL]

I think that SSD's are a very good thing for HD's over the next few years: There has been a significant complacency among HD manufacturer's for quite a while. There hasn't been any driving need to go beyond 1TB because most people just don't need that much local storage. Further, most people seem to be happy with the speeds of the drives, so that if anything the manufacturers have been slowing down the speeds to cheapen the manufacturing costs: A good example is the WD 1TB drive.

I would argue, that because of the threat of SSD's the HD manufacturer is now going to have to change their focus. I am absolutely sure, that they do not wish to be relegated to tape-drive status. However, if SSD's start replacing HD's as the standard storage HD's manufacturers will embrace the only market left to them and compete on storage capacity and cost/GB.

Meanwhile, to prevent and/or delay the loss of being the main storage device the companies will start innovating to speed HD's up so as to compete with SSD's for as long as they can. Also, they will start to increase capacity and cost/GB so that if they do end up just being a backup media there will be a market to fall back upon. The capacity and cost/GB should be easy just using the promise of perpendicular recording -- There just hasn't been the need to drive them to create new products. Speeding up the HD, is going to be harder but is definitely possible especially as storage capacity/mm^2 improves: Wouldn't it be neat to have a drive so dense that there is no seek time but only rotational latency. I just don't know if they can speed up the drive speeds fast enough to limit the damage SSD's are going to do to their market share, but they are going to have to try..

In the longer run, I would be worrying as a HD manufacturer because if they can't find a technological/market solution to the threat of SSD's then they will become the floppy of the 21st century.

 

[jtr1962]

In the longer run, I would be worrying as a HD manufacturer because if they can't find a technological/market solution to the threat of SSD's then they will become the floppy of the 21st century.

I think it's going to be a case of if you can't beat them/join them. Mechanical disks will never be able to compete with SSDs in terms of latency. There's just no way something which rotates can have a sub-0.1 ms access time. For now they can compete in transfer speeds, but even that advantage is slowly being eroded, with the potential of SSDs to be much faster there as well. For power usage, something which is becoming increasingly important, HDDs fail badly as well. Speeding them up to reduce access times will only make them consume more power. In the end cost per GB and raw capacity are the only reasons mechanical disks will survive. In both areas SSDs are gaining exponentially while HDDs are only gaining linearly. Unless there's some major breakthroughs, I don't see the capacity of a 3.5" mechanical disk getting much past a few TB, ever. SSD on the other hand also may have limits, but for now they're not as big an issue.

Honestly, if I were any of the main manufacturers of hard disks I would start doing heavy research into various types of solid-state storage. Looking out much past ten years, perhaps even five, I really don't see much of a future for any type of spinning disk. To draw on the imperfect analogy using LEDs, LEDs just have the potential to be so much better than what they're replacing (incandescent and eventually fluorescent) that the other technologies just don't have a prayer. Like mechanical disks, the only thing currently keeping the other light sources alive is cost per lumen. Once LED gets close in that area, they're gone. Same thing for mechanical disks.

 

[ddrueding]

I really don't see much of a future for any type of spinning disk.

That got me thinking. What is the price differential between 512MB thumbdrives and CD-R media? Will software take this intermediary step, or go entirely to online distribution?

 

[Will Rickards]

I'm not sure I agree that HD companies are complacent. I think the increased density and perpendicular recording were significant hurdles requiring quite a bit of research.

I also think 1TB is not that significant an amount of space. It is probably plenty for now for the average computer user. But as video gets more hi-def and so do pictures the storage requirements go up and up. It certainly isn't that significant an amount of space for a business. They need lots and lots of TB for storing data for users.

I do think it is a sweet spot right now of just the right size for a single hard drive. It is still expensive at $300 per drive, though if you put it in historical perspective that is not really that expensive.

As far as SSD's, I can't wait for them honestly but the average home user probably doesn't care very much. And when he sees a high price for getting one with his shiny new laptop (it is +900 I think for the samsung one) he probably won't bite. But the rise of the laptop versus the desktop may mean SSD's will catch on earlier than we think.

But they really have a capacity problem. 32GB or 64GB or even 128GB is really not that much.

I think a combination drive might be best and in the works. Something that uses say 32GB of flash for a cache and has 1TB of storage. Not just as a replacement for the normal cache of a hard drive but something more proactive. It would store all your commonly used files on the SSD as well as the HD and page in as necessary.

 

[Pradeep]

From what I understand 20K SAS drives will be with us "soon"

I think there will definitely come a time when solid state exceeds spinning disks, it's just a matter of time as process shrinks occur and capacities increase. Having a rotating mechanical device that is bulky and liable to fail is an episode in the data evolution.

The biggest requirement in the enterprise is IOs per sec. Currently a 15K SAS/FC drive can provide at best 180 IO/sec. To get decent performance for a large mail system you may need 30-40 spindles, including RAID 6 redundancy. That's three + sleds of disk, 14 drives in each. You don't want to know how much it costs. If a single solid state drive can deliver 500GB and 20,000 IO/sec, SAN performance is going through the roof. Just what we need to keep the quad cores (8-16 to come down the road?) fed with data. And what starts in the enterprise will soon be at consumer levels.

Tape drives are alive and well, keeping up with the increases in hard drive density. Biggest problem is keeping them fed with data. We ended up going with LTO3 because the drives and media were cheaper. Backing up snapshots on a SAN allows them to run at rated speed, rather than wait on a network connection. For normal ethernet backup streams LTO4 is a little too fast at 120MB/sec, starts having to slow the tape. What else to do you store archives/off site backups on?

 

[ddrueding]

What else to do you store archives/off site backups on?

I just use more drives.